Skip Links

USGS - science for a changing world

Techniques and Methods 11-C3

U.S. Geological Survey Center of Excellence for Geospatial Information Science (CEGIS)

Estimating Prediction Uncertainty from Geographical Information System Raster Processing: A User’s Manual for the Raster Error Propagation Tool (REPTool)

By Jason J. Gurdak, Sharon L. Qi, and Michael L. Geisler


Thumbnail of and link to report PDF (7.0 MB)

The U.S. Geological Survey Raster Error Propagation Tool (REPTool) is a custom tool for use with the Environmental System Research Institute (ESRI) ArcGIS Desktop application to estimate error propagation and prediction uncertainty in raster processing operations and geospatial modeling. REPTool is designed to introduce concepts of error and uncertainty in geospatial data and modeling and provide users of ArcGIS Desktop a geoprocessing tool and methodology to consider how error affects geospatial model output. Similar to other geoprocessing tools available in ArcGIS Desktop, REPTool can be run from a dialog window, from the ArcMap command line, or from a Python script.

REPTool consists of public-domain, Python-based packages that implement Latin Hypercube Sampling within a probabilistic framework to track error propagation in geospatial models and quantitatively estimate the uncertainty of the model output. Users may specify error for each input raster or model coefficient represented in the geospatial model. The error for the input rasters may be specified as either spatially invariant or spatially variable across the spatial domain. Users may specify model output as a distribution of uncertainty for each raster cell. REPTool uses the Relative Variance Contribution method to quantify the relative error contribution from the two primary components in the geospatial model—errors in the model input data and coefficients of the model variables.

REPTool is appropriate for many types of geospatial processing operations, modeling applications, and related research questions, including applications that consider spatially invariant or spatially variable error in geospatial data.

First posted July 23, 2009

For additional information contact:

USGS Colorado Water Science Center
Box 25046, MS 415
Denver, CO 80225

Or visit the Colorado Water Science
Center Web site at:

Part or all of this report is presented in Portable Document Format (PDF); the latest version of Adobe Reader or similar software is required to view it. Download the latest version of Adobe Reader, free of charge.

Suggested citation:

Gurdak, J.J., Qi, S.L., and Geisler, M.L., 2009, Estimating prediction uncertainty from geographical information system raster processing—A user’s manual for the Raster Error Propagation Tool (REPTool): U.S. Geological Survey Techniques and Methods 11–C3, 71 p.




Purpose and Scope

Overview of Raster Error Propagation Tool (REPTool)

Getting Started

System Requirements

Installation and Execution of REPTool

Program Capabilities and Characteristics

Program Limitations

Raster Processing, Error, and Uncertainty

Theory of Error Propagation and Uncertainty

Quantitative Error Model

Assigning Error

Application of Quantitative Error Model and Error Propagation in GIS

Latin Hypercube Sampling Method

REPTool User’s Guide

Input Instructions

Preparing Rasters for REPTool

Input Rasters and Errors

Model Coefficients

Distribution Type

Model Equation

How to Write Model Equations in REPTool

Number of Iterations

Output Percentiles and Workspace

Advanced Parameters

Description of Output Files

Example Problem


References Cited


Appendix 1–Statistical Functions

Algorithm to Compute Inverse Normal Cumulative Distribution Function

Normal Cumulative Distribution Function

Lognormal Cumulative Distribution Function

Uniform Cumulative Distribution Function

Appendix 2–Command-Line Syntax and Python Scripting

Command-Line Syntax

Command-Line Example

Scripting Syntax

Script Example

Appendix 3 – Developer Documentation



How to Read Developer Documentation


Package Architecture: CEGIS_001

Package Architecture: control

Package Architecture: datasources

Development Test-Bed Packages

Package Architecture: main

Package Architecture: reptool

Package Architecture: runnables 1 of 2

Package Architecture: runnables 2 of 2

Package Architecture: Virtual Machine (VM) 1 of 6

Package Architecture: Virtual Machine (VM) 2 of 6

Package Architecture: Virtual Machine (VM) 3 of 6

Package Architecture: Virtual Machine (VM) 4 of 6

Package Architecture: Virtual Machine (VM) 5 of 6

Package Architecture: Virtual Machine (VM) 6 of 6

Glossary for Developer Documentation

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: GS Pubs Web Contact
Page Last Modified: Friday, 02-Dec-2016 15:49:07 EST